This study presents the development of a three-dimensional disaster visualization system designed to support intuitive evacuation planning in urban environments. Targeting the city of Yokohama, the system integrates tsunami and flood inundation data with designated emergency shelter information using Cesium for Unity and GeoJSON datasets. High-resolution 3D terrain and building models are streamed from the cloud, enabling lightweight performance while maintaining spatial accuracy. The system features a glasses-free stereoscopic display (ELF-SR1), allowing users to perceive depth and spatial relationships intuitively. By using a gamepad interface with altitude-sensitive camera controls, users can explore complex urban structures and evaluate evacuation routes interactively. Performance evaluation in the Minato Mirai district demonstrated fast load times and high frame rates, confirming the system's feasibility for real-time disaster risk communication. Future extensions will include user evaluation and use-case analysis to further validate the system's feasibility and applicability in real-world disaster preparedness. This platform contributes to improving spatial awareness and fostering proactive disaster preparedness through immersive geospatial visualization.

This study develops an innovative 3D crime prevention map system for Yokohama City, Japan, integrating Google’s Photorealistic 3D Tiles with local crime statistics. Utilizing Unity and Cesium for Unity, the system visualizes crime patterns in a 3D urban context, revealing correlations between urban morphology and criminal activity. Findings indicate higher crime rates in commercial districts with complex alley structures, while areas with enhanced visibility show lower incidence. This novel approach offers valuable insights for urban safety enhancement and crime prevention strategies, potentially influencing future urban planning methodologies.

This study explores the integration of generative AI in game development, focusing on shooting games. We propose a hybrid approach combining AI capabilities with human creativity, utilizing the PLATEAU 3D city model and spatial reproduction displays. Our method addresses AI limitations in creative tasks by imposing technological constraints, resulting in innovative gameplay experiences. We implemented generative AI across various development stages, including character design and programming. The integration of PLATEAU data and spatial reproduction displays within Unity created an immersive stereoscopic environment. While AI enhanced development efficiency, limitations were observed in creative tasks and novel device programming. Our findings highlight the potential of generative AI in game development and suggest applications beyond entertainment. This study provides a foundation for effective AI utilization in game creation, emphasizing the importance of a balanced hybrid approach.

In cities, outdoor advertisements are a major spatial component that attracts attention, and the effectiveness of outdoor advertisements has been measured using users' gaze. There are several eye-tracking devices, but they are not easy to use because they require special and expensive devices. In this study, we developed an eye tracking tool using a smartphone camera to evaluate outdoor advertisements. The developed system is capable of recording video of gazing points in a real urban space. This system has a wide range of applications, including new advertising contents, landscape evaluation, and driving support.

The video posting service is a web service that is familiar to many people and is expected to be in demand in the future. In the video posting site Nico Nico Douga, we proposed an interface in which user comments are superimposed on the video and played on the screen. In this study, we examined a new video posting service that allows users to post their comments not as text, but as audio, and verified the possibility of such a service. As a result, through the prototyping of the system, it became clear that it has the potential to become an unprecedented new video posting service. On the other hand, the proposed system has some problems, such as the difficulty of cancelling voice comments afterwards. In order to solve these problems, we need to devise a better operation method and improve the system. In the future, we will improve these issues through user evaluation.

Automatic doors use sensors to open and close the door, so the door opens and closes only when it recognizes a person. The problem is that the door opens and closes when a person who has no intention of opening the door passes by the automatic door. In this study, we propose an automatic door with loose security by using recorded words and poses. By developing an automatic door control system that only allows people who know the specific words and poses to pass through, we can maintain a certain level of security and avoid unnecessary opening and closing motions.

In this study, we developed a soundscape system that reproduces a real soundscape in 3D in a virtual space, based on the "100 Best Soundscapes in Japan" selected by the Ministry of the Environment. Using the game engine "Unity", we proposed a method of embedding recorded sound sources into virtual objects placed in a virtual space, and sequentially playing back the sound when the user walks or turns around in the virtual space. By using this system, it is possible to reproduce the same sound field in the virtual space as in the real space, and it can be applied in various places.

During a child's growth process, doll play contributes to the formation of self, creativity, and social development. In common dollhouse play, children move the dolls and imagine the world of the dollhouse. In this study, distance and pressure sensors were installed in the dollhouse and projection mapping was used to project interactive images in response to the measurements. The images were generated in real-time in response to doll and hand movements, which were then reflected in the dollhouse, allowing for a wider range of doll play.

In this study, we developed a character control interface for a virtual character using fuzzy voice input in Japanese. Previous voice input interfaces have used pre-registered words in the system to control characters. The system uses the Japanese WordNet, a Japanese semantic dictionary, to associate words including synonyms with the character's commands so that even fuzzy voice input can be easily controlled. This system is considered to be effective as a voice input interface for systems that require instantaneous operations, such as action games.

Herein, we present a landscape simulation system that reproduces general town scenery from the past using a game engine to understand the historical culture of a given area. To reproduce a general townscape of the Edo era, we referred to historic documents and maps, and sought feedback from the local residents to develop the system. In addition, because we use a game engine, our proposed system that can change the time zone, season, and weather as specific expressions. Furthermore, we attempted Ukiyo-e style rendering. Some residents evaluated the system using head-mounted display. We found that the components modeled by the proposed system were accurate. The developed system will be opened to all local residents shortly

Hazard maps are currently developed using computationally expensive techniques and must be revised after a major disaster. Therefore, we developed a low-cost tsunami evacuation simulation system using a game engine and open data. A simulation evaluation was conducted using the target area of Kamakura City, Japan. We also developed an agent that performs evacuation actions and autonomously searches for evacuation destinations and evacuation behaviors at specified speeds in order to clarify current issues with the location of evacuation sites. The agent prepared three walking speeds, three disaster conditions, and two evacuation behaviors and could freely change the number and ratio of agents. The simulation results revealed locations with high concentrations of tsunami victims, even in inland areas.

At present, hazard maps are developed using computationally expensive techniques. After a major disaster, these maps must be revised. We developed a tsunami evacuation simulation system using a game engine and open data at low cost. An agent that performs evacuation actions, and autonomously searches for evacuation destinations and evacuation behaviors at specified speeds was developed, to clarify current issues with the location of evacuation sites. The agent prepared three types of walking speeds and disaster conditions, and two types of evacuation behaviors, and could freely change the number and ratio of agents. As a result of the simulation, it became clear that there are places where victims are concentrated even in inland areas.

In recent years, digital contents are attracting attention again, but in the past many passive types of exhibits are present, and it is considered that giving interactive ability is important. In this research, we propose a production method assuming use such as exhibition. Using the produced aquarium type device, two types of images are projected from different directions onto underwater film and combined. One is a motion object detected by a camera, an image with an effect applied in real time, and the other is an image giving an effect based on user's physical information. With this method, it is possible to make the two types of images feel the depth of the space, and it is possible to create a new dynamic presentation. We also conducted an evaluation study on the color of the image to be projected and examined the images to be used.